Today, COVID-19 is one of the most severe issues that people are grappling with. Half of the faces are hidden by the mask in this instance. The region around the eyes is usually the sole apparent attribute that can be used as a biometric in these circumstances. In the event of a pandemic, the three primary biometric modalities (facial, fingerprint, and iris), which commonly enable these tasks, confront particular obstacles. One option that can improve accuracy, ease-of-use, and safety is periocular recognition. Several periocular biometric detection methods have been developed previously. As a result, periocular recognition remains a difficult task. To overcome the problem, several algorithms based on CNN have been implemented. This chapter investigated the periocular region recognitions algorithms, datasets, and texture descriptors. This chapter also discuss the current COVID-19 situation to unmask the masked faces in particular.
TopIntroduction
Biometrics is a field that looks at a person's biological characteristics that are unique. Biometrics methods used to determine the distinctive behavioural and physical characteristics of humans. It is the preferred means of identification, outperforming traditional methods such as passwords and PINs (G. Liu, at al, June 1997). Biometric devices aid biometric system authentication and identification by utilising a number of unique human attributes such as fingerprints, vein patterns, DNA sequencing, hand geometry, iris pattern, voice pattern, face detection, and signature dynamics (Kumari P, Seeja KR., 6 June 2019). The fingerprint, which is a physical characteristic, is the first and most common thing that comes to mind when discussing distinct features (U.J. Gelinas,at al, 2004) (N. Osifchin and G. Vau, 1997). In our day-to-day uses for biometrics as a way of authentication, there are a variety of biometric traits that can be utilized to identify humans (K. Kimura and A. Lipeles, 1996). Biometric-based authentication is more secure than any other technique since it binds an identity to a specific person rather than a password or a code that anybody could use.
One of the most fundamental aspects in facial computation is the classification of certain facial traits from photos and videos. (https://www.koreatimes.co.kr/www/nation/2019/01/371_262460.html,2019).These significant traits, such as the distance between the eyes and the relative placements of the nose, chin, and mouth, are combined to generate a facial profile that aids in the identification of each individual.
Face biometrics stand out among other biometrics because they do not demand an individual's active engagement (H. Zhang, 1997). Many scientists are interested in face recognition, and as a result, it has become a gold standard in the field of human recognition. It has been the most exhaustively investigated field in computer vision for more than four decades (J. M. Smereka and B. V. K. V. Kumar, 2013). Face recognition is a widely used biometric that works well in a controlled environment. Face recognition systems, on the other hand, perform worse when the face is partially obscured. Surveillance videos frequently do not show the entire face of crooks. Helmets, hair, glasses, and skiing masks are used to conceal the face in various settings. Furthermore, women in other nations conceal their faces partially owing to cultural and religious reasons. Face recognition suffers from a loss of accuracy and reliability when persons wear surgical masks (L. Bass,at al, 2003), despite being unexpectedly accurate and reliable in the presence of partial facial occlusions.